On 01/11/2011 11:58 AM, Eric Blake wrote:
>> char *
>> openat_proc_name (char buf[OPENAT_BUFFER_SIZE], int fd, char const *file)
>> {
>> @@ -98,7 +99,13 @@ openat_proc_name (char buf[OPENAT_BUFFER_SIZE], int fd,
>> char const *file)
>> else
>> {
>> size_t bufsize = PROC_SELF_FD_NAME_SIZE_BOUND (strlen (file));
>> - char *result = (bufsize < OPENAT_BUFFER_SIZE ? buf : xmalloc
>> (bufsize));
>> + char *result = buf;
>> + if (OPENAT_BUFFER_SIZE < bufsize)
>> + {
>> + result = malloc (bufsize);
>> + if (! result)
>> + return NULL;
>
> Either we must depend on malloc-posix, or you need to force errno=ENOMEM
> here rather than relying on malloc() to do it.Correction - due to how openat_proc_name is used, we _don't_ need malloc-posix, because you didn't change things to promise a sane errno. Since we _always_ fallback to the fchdir method if this failed without ever inspecting errno, the caller never needs to know that we had an intermediate ENOMEM situation, and we can get by without worrying whether malloc set errno. But you should still drop the xmalloc dependency from the module description. -- Eric Blake ebl...@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
